I did figure out what is going wrong, but I don't know how to fix it. Let me start by showing the tail end of the page creation process, where this thing is going off the rails:

first value is /home/bob/2.scripts/pages/3.cw/template_stuff/aimages 
Use of uninitialized value in concatenation (.) or string at template_
+stuff/html7.pm line 127, <STDIN> line 2.
array part is 
undefined!...initializing:
a is /home/bob/2.scripts/pages/3.cw/template_stuff/aimages/quux
b is quux
AoA put failed 
first value is /home/bob/2.scripts/pages/3.cw/template_stuff/aimages 
Use of uninitialized value in concatenation (.) or string at template_
+stuff/html7.pm line 127, <STDIN> line 2.
array part is 
undefined!...initializing:
a is /home/bob/2.scripts/pages/3.cw/template_stuff/aimages/quux
b is quux
AoA put failed 
return is 3.cw15.html
$
[download]

The offenders are create_page and put_page, who until recently, were part of main. Relegating them to be subroutines was great for legibility in main, but I seem to have developed issues. First of all, even though it tells me it fails every time, it gets a page on the net. The resulting page is weirdly-populated. ugly page

My main data struture is a hash that I pass into all my subroutines:

  my $rvars = shift;
  my %vars  = %$rvars;
[download]

But then I want to create something with persistence, like:

  $vars{refc} = get_content($rvars);
  $rvars = \%vars;  ## will same trick work?
[download]

and then

  print "Put file to server(y/n)?: ";
  my $prompt2 = <STDIN>;
  chomp $prompt2;
  if ( $prompt2 eq ( "y" | "Y" ) ) {
    put_page( $sftp, $rvars );
  }
[download]

$ref_content's value is NULL, so this is not an effective pass.

  my $ref_content = $vars{refc};
  my @AoA         = @$ref_content;
  say "content----------";
  print Dumper $ref_content;
  for my $i ( 0 .. $#AoA ) {
    say "first value is $vars{to_images} ";
    say "array part is $AoA[$i][0]";
    if (!defined $AoA[$i][0]) {
    say "undefined!...initializing:";
    $AoA[$i][0] = 'quux';
}
    my $a = path( $vars{to_images}, $AoA[$i][0] );
    say "a is $a";
    my $b = $a->basename;
    say "b is $b";
    $sftp->put( $a, $b ) or warn "AoA put failed $@\n";
  }
[download]

It didn't always used to be this broken, so I'm happy to entertain suggestions....

Ok, you've confirmed that $vars{refc}[$i][0] is undef, but as I said, please use Data::Dumper (or Data::Dump) to show the contents of $vars{refc}. Probably the data structure that get_content is returning doesn't match what you're doing with it in put_page.

please use Data::Dumper (or Data::Dump) to show the contents of $vars{refc}

What it showed was that the content was populated. What was causing the entire template to go wonky was that there were more files in rus_captions than in eng_captions or images. I filled undefined things with bubble gum and just got them on the net. Most of the complaining is about not being able to sftp non-existent files. I wanted to come up with some baseline defaults so that the whole page doesn't go down for lack of an image to stand in for a while.

I've gotten rid of a lot of mistakes as I dug my way back into this one. The first was not having brackets around my qw in the initialization of the array:

elt 0 0 is й
elt 0 1 is  
elt 0 2 is  
elt 0 3 is л
elt 0 4 is  
elt 0 5 is  
elt 0 6 is с
elt 0 7 is  
elt 0 8 is  
Can't use string ("о") as an ARRAY ref while "strict refs" in us
+e at template_stuff/utils1.pm line 61.
$
[download]

I don't see a way to use Dumper and have that the be cyrillic, because that's what I want to see. I use Dumper for doing anything to get on the scoreboard. (Not such a high price, except that you condemn yourself to read a lot of your own data: uggh)

I've been trying to write a better print_aoa_utf8 but have only partial success. Partially, my data set needed more conditioning. The hardest thing for me to represent like this is a space.

in make russian xword------
elt 0 0 is й
elt 0 1 is  
elt 0 2 is  
elt 0 3 is л
elt 0 4 is  
elt 0 5 is  
elt 0 6 is с
elt 0 7 is  
elt 0 8 is  
elt 1 0 is о
elt 1 1 is б
elt 1 2 is л
elt 1 3 is о
elt 1 4 is м
elt 1 5 is о
elt 1 6 is в
elt 1 7 is '
elt 1 8 is '
elt 1 9 is ш
elt 2 0 is 1
elt 2 1 is 2
elt 2 2 is 3
elt 2 3 is 4
elt 2 4 is 5
elt 2 5 is 6
elt 2 6 is 7
elt 2 7 is 8
elt 2 8 is 9
return1 is nothing nyet

What I have is:

sub print_aoa_utf8 {
  use warnings;
  use 5.011;
  use utf8;    # a la François
  use open OUT => ':encoding(utf8)';
  use open ':std';

  my $a   = shift;
  my @AoA = @$a;

  for my $i ( 0 .. $#AoA ) {
    my $aref = $AoA[$i];
    for my $j ( 0 .. $#{$aref} ) {
      print "elt $i $j is $AoA[$i][$j]\n";
    }
  }
  return $a;
}
[download]

With my difficulties getting any values initialized. I wonder how I might make it a matter of course and think that a toolkit solution would be a reasonable goal, in particular because I want to have a means for considering fonts, and we saw a tk version of trying out different fonts not long ago.

I had to make sure that all previous russian captions were erased before I went and got more:

  print "Make rus captions(y/n)?: ";
  my $prompt1 = <STDIN>;
  chomp $prompt1;
  if ( $prompt1 eq ( "y" | "Y" ) ) {
    ## delete existing files
    foreach my $child ( $vars{rus_captions}->children ) {

      my $success = $child->remove;
      say "success deleting was $success";
    }
    my $ref_cap = make_russian_captions($rvars);
  }
[download]

STDOUT then shows:

Make rus captions(y/n)?: y
success deleting was 1
success deleting was 1
success deleting was 1
success deleting was 1
matching are a.txt b.txt
rus_munge is /home/bob/2.scripts/pages/7.cw/template_stuff/translation
+s/trans.07-01-2019-16-15-08.txt
rus_path is /home/bob/2.scripts/pages/7.cw/template_stuff/ruscaptions/
+a.txt
rus_path is /home/bob/2.scripts/pages/7.cw/template_stuff/ruscaptions/
+b.txt
[download]

And I figured all this out by looking at the data as haukex suggested. I've had a nice few days coding that reduces the ugly, but not completely.

sub make_russian_crossword {
  use 5.011;
  use warnings;
  use POSIX qw(strftime);
  use Path::Tiny;
  use Encode;
  use open OUT => ':encoding(UTF-8)', ':std';
  use Data::Dumper;

  my $rvars = shift;
  my %vars  = %$rvars;

  my $munge = strftime( "p%d-%m-%Y-%H-%M-%S\.txt", localtime);
  my $in_path = path( $vars{rus_captions}, $munge )->touchpath;
  
  ##Let mother know that you created a file, *verb* a reference:
  $vars{log_file}=$in_path;
  $rvars = \%vars;   #does this line do anything for me?
 

  # open file for writing
  # my $fh = path($in_path)->openw_utf8; not helping
  say "in make russian xword------";
  my $data = $vars{cw_data};
  #print Dumper $data;
  print_aoa_utf8($data);
  #print $fh $data;
  my $a   = $data;
  my @AoA = @$a;
  for my $i ( 0 .. $#AoA ) {
    my $aref = $AoA[$i];
    for my $j ( 0 .. $#{$aref} ) {
       "$AoA[$i][$j]";
    }
  print "\n";
  }
  # ->append_utf8(@data);
  return "nothing nyet"
}
[download]

Does the 3rd line here do anything for me or the program?

  ##Let mother know that you created a file, *verb* a reference:
  $vars{log_file}=$in_path;
  $rvars = \%vars;   #does this line do anything for me?
[download]

What I'm reaching for at the end is a control that will go over the array, which can be seen as a column vector of row vectors. I want to iterate over the columns and print out one row at time, with no spaces and a newline at the end. Fishing for notation as to how to do that. With the output as a series of outputs of row vectors, then I think it will fit in well with the Path::Tiny method call append_utf8(@data). Then we can see how these things look on the net. Right now, it's looking pretty stable: crossword prototype shaping up.

Again, very grateful for comments,